ROC analysis

The ROC analysis presented here was performed on the data including all groups (22q, CR, ZR, NC). Groups status was coded as “22q” or “not22q”.

Sample sizes:

22q CR NC SZ
168 19 97 41

ROC curve

ROC curve with optimal cut-off point (Specificity, Sensitivity). Optimal cut-off is defined as the point that maximizes Youden’s index defined as \(J = sensitivity + specificity − 1\)

## Area under the curve: 0.8845
## 95% CI: 0.8495-0.9195 (DeLong)

PCA

PCA was performed without scaling gestalt scores (covariance PCA).

Whole sample

Scree

Factor scores

Parallel Analysis

Needs work.

PC1

Loadings

ROC Analysis

## Area under the curve: 0.9046
## 95% CI: 0.873-0.9361 (DeLong)
Compare ROCs: 22q Gestalt score vs PC1 as input

## 
##  DeLong's test for two correlated ROC curves
## 
## data:  resROC and resROC_PCA
## Z = -2.7122, p-value = 0.006684
## alternative hypothesis: true difference in AUC is not equal to 0
## sample estimates:
## AUC of roc1 AUC of roc2 
##   0.8845352   0.9045723

A weighted combination of syndrome scores improves the AUC relative to predicting 22q status based on only 22q syndrome scores.

PC2

Loadings

ROC Analysis

## Area under the curve: 0.6753
## 95% CI: 0.5894-0.7611 (DeLong)
T-test: Normal Controls vs Schizophrenia
## 
##  Welch Two Sample t-test
## 
## data:  NC_score and SZ_score
## t = 3.7495, df = 83.679, p-value = 0.0003253
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.07023095 0.22887998
## sample estimates:
##   mean of x   mean of y 
##  0.13597450 -0.01358097
## Cohen's d |       95% CI
## ------------------------
## 0.67      | [0.29, 1.04]
## 
## - Estimated using pooled SD.
T-test: Normal Controls vs Clinical Risk
## 
##  Welch Two Sample t-test
## 
## data:  NC_score and CR_score
## t = 3.8668, df = 33.034, p-value = 0.0004897
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.08249714 0.26569048
## sample estimates:
##   mean of x   mean of y 
##  0.13597450 -0.03811931
## Cohen's d |       95% CI
## ------------------------
## 0.78      | [0.28, 1.28]
## 
## - Estimated using pooled SD.
T-test: Schizophrenia vs Clinical Risk
## 
##  Welch Two Sample t-test
## 
## data:  SZ_score and CR_score
## t = 0.48897, df = 42.756, p-value = 0.6274
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.07668375  0.12576044
## sample estimates:
##   mean of x   mean of y 
## -0.01358097 -0.03811931
## Cohen's d |        95% CI
## -------------------------
## 0.13      | [-0.42, 0.67]
## 
## - Estimated using pooled SD.

PC3

Loadings

PC4

Loadings

Group means

22q group

Scree

Loadings

PC1

PC2

PC3

PC4